Microsoft Imposes Internal Limits on AI Token Usage to Control Costs

Microsoft has introduced strict budget limits on the internal use of artificial intelligence tokens for its engineers. According to an internal email obtained by 404 Media, the company is taking these steps to curb excessive computational resource spending. As part of this initiative, Microsoft has transitioned its internal tools to GPT-5.6, a version designed to be more economical and efficient. Employees now have access to a tracking dashboard to monitor their own spending, and the company warned that further restrictions may be applied if consumption remains uncontrolled.

This move comes amid a broader industry trend where companies are struggling with the high costs of AI. The practice of tokenmaxxing, where developers use AI to generate massive volumes of code without proportional productivity gains, has led to some engineers spending thousands of dollars per month. Other major corporations, including Amazon, Adobe, Atlassian, Citi, Uber, and Meta, have also implemented restrictions due to high costs. An executive from Nvidia has noted that AI can be more expensive than maintaining human employees.

Despite these restrictions, Microsoft's recent financial reports show growth in revenue and profits that exceeded Wall Street estimates. The current shift is viewed as a correction in response to investor criticism regarding the massive spending by big tech firms on AI.